The First Wearable AI Heart Monitor Has Arrived, and It’s Saving Lives

Health

The new Heartsense heart monitor is the first of many wearable AI-driven devices that will help reshape healthcare in the near future.

How can artificial intelligence (AI) help you stay on top of your heart health? A new device hopes to answer that.

With a new high-tech, wearable heart monitor, Dr. Rameen Shakur is hoping that AI will help blast cardiology into the future.

In the first quarter of 2019, Shakur will introduce the Heartsense monitor — billed as the first AI-driven wearable heart monitor — to the market.

The monitor is being released by Cambridge Heartwear, the Cambridge, United Kingdom–based company that Shakur co-founded in 2017.

For Shakur, this device occupies an important niche that he noticed other monitors weren’t necessarily filling.

“You’d see people walk around with these monitors that never understood the patient experience — these Holter devices that feel like having an octopus stuck to your old ’80s Walkman that’s strapped to your side,” Shakur, currently a fellow at the Massachusetts Institute of Technology (MIT), told Healthline. “It got me thinking, ‘This is exactly why the majority of our patients don’t keep monitors on long enough. They’re not ergonomic, they’re not comfortable, they don’t fit in with daily life.’”

He added, “In the U.K., unfortunately, one has to go through their primary care who refers you to get a device. It’s like a four-to-six-week wait before you get a monitor. In that time, it’s like a ticking time bomb; you could get a stroke in that time period. I kept thinking, ‘How can we really get people diagnoses in real time if they have to go through this process to get these monitors?’”

The birth of the Heartsense monitor

He decided to start the process of developing a more efficient and accessible device himself.

In 2015, Shakur met Roberto Cipolla, a professor at Cambridge University’s engineering department, whose father had died of a stroke the year before. The two decided to collaborate on devising the monitor. They officially launched their company in 2017.

“I have no previous electric engineering background, no technology experience. I started working with colleagues in artificial intelligence and engineering because I wanted a way for people who used the monitor to get their data in real time, immediately,” he said.

The resulting Heartsense monitor looks like a simple band worn around the chest.

Waterproof, the device includes multiple sensors that take electrocardiograms (ECG), or recordings of your heart’s electrical activity.

These recordings are used to determine your heart rhythm and to pinpoint any irregularities. Data collected from these readings are immediately streamed to cloud storage, which is where the AI comes in.

AI algorithms pinpoint irregular rhythms, delivering data to you immediately through a mobile app that you can share with your physician at your next appointment.

The field is hot

The need for the cardiology field to continue to perfect heart monitoring practices is great.

In the United States alone, strokes kill 140,000 people each year, and someone has one every 40 seconds. Additionally, every four minutes, someone dies from a stroke, according to the Centers for Disease Control and Prevention (CDC).

Mobile heart monitors like this one are particularly useful in spotting atrial fibrillation, or AFib, which is the most common type of heart arrhythmia, or irregularity.

When someone has AFib, the regular beating in the heart’s upper chambers, or atria, is irregular, resulting in blood not flowing normally down to the lower chambers, which are known as the ventricles.

The CDC says this affects 2.7 to 6.1 million people in the United States.

Each year, AFib leads to 750,000 hospitalizations, nationwide, and costs the country a whopping $6 billion annually.

Given how widespread this is, obviously, the market to bring easy heart monitoring to the masses is pretty full.

Cleveland Clinic cardiologist Dr. Dan Cantillon says “the field is hot.”

“Currently, the most widely utilized devices are the AliveCor mobile and watch band, and of course the newest Apple Watch. In general, ECG monitoring devices apply detection algorithms of varying accuracy that are continually refined and updated to improve their performance. The major difference with machine learning (ML) applications is that the software code essentially updates and refines itself to improve performance,” Cantillon told Healthline. “While there’s greater capacity to learn and improve rapidly, the disadvantage is a loss of understanding of how it works exactly, depending on the ML methodology. Whereas even very sophisticated human-created algorithms can be understood by the principles of logic and mathematics. But again, the human computing capacity cannot even remotely match that of the machine.”

However, that doesn’t mean machines are replacing human expertise.

Cantillon said that there a few concerns people should have when it comes to these kinds of devices. One involves telling apart an accurate ECG signal from “noise.”

“These devices have to filter out noise while often amplifying the true ECG signal,” he added. “Technical aspects like sampling frequency, and filters really matter in addition to simply getting a good recording where the sensing electrodes make contact with the skin. Simply put, ‘garbage in’ will result in ‘garbage out’ regardless of how good the software performs.”

In addition to this concern, Cantillon explained that it is important that all of the new wearable heart monitoring technology that is emerging right now needs to go through “truly robust clinical validation testing.”

He said that human algorithms and machine learning tools are vulnerable to various kinds of bias — “they can only learn from what they’re given.”

“There are plenty of examples where machine learning can be unknowingly biased,” Cantillon said. “For example, ML photo recognition software trained to tell the difference between a dog and a wolf can get it wrong by considering snow in the background. In cardiology, robust testing is needed for patients across an entire spectrum of perfectly healthy people to very sick patients with heart disease. Validating with hundreds of patients is probably not enough. Truly robust systems will ultimately need to validate with much larger sample sizes and complexity.”

Cantillon added that start-up companies can lack the resources to carry out this kind of testing. Given how eager many of these businesses are to beat the competition to hit the market, sometimes products can hastily make it out there, with start-up founders thinking that they can “expand validation testing later on,” or that it will eventually happen if the start-up is acquired by a larger company.

For its part, in perfecting the Heartsense monitor, Cambridge Heartwear carried out clinical trials with primary care patients enrolled in Lancashire, United Kingdom.

Right now, the monitor is also going through clinical trials with athletes in both the United Kingdom and the United States.

Shakur said one thing that sets the monitor apart from competitors is the fact that it took into account the different physiologies of male and female consumers. He said one complaint he has heard in the past is that other monitors might not always fit comfortably for female bodies.

“When we were doing some of the prototype testing of the device, we were surprised how the medical device community misses 50 percent of the population by not considering the female body form,” he said. “Most devices, if not all, have forgotten that there are certain anatomical differences between males and females. This device address these issues, so that, women, in general, who want to have these devices can go about their daily lives without being in discomfort.”

The way of the future

Dr. Gordon Tomaselli, FAHA, FACC, FHRS, the Marilyn and Stanley M. Katz Dean at Albert Einstein College of Medicine, said that we will only continue to see more AI and machine learning integrated into cardiology in the future.

“I think that we’re really beginning to scratch the surface of what AI can do in medicine. It’s going to have a big impact. We keep telling our students, in fact, they do not necessarily have to be data scientists and engineers but they’ll have to be able speak to them. We tell them that this is the kind of information they’ll have to integrate into their clinical practices, not by themselves, but with the assistance of people doing this kind of work,” Tomaselli, who is also a spokesperson for the American Heart Association, told Healthline. “This will improve physician efficiency and will be used to address issues of population health and how we deploy therapies to patients.”

He said high-tech methods will increasingly provide the backbone for “precision medicine,” as hospitals become increasingly savvy at leveraging big data to provide the most targeted treatments possible for a range of conditions.

For example, in 2017 IBM announced partnerships with hospitals to use its so-called “supercomputer” Watson — yes, the AI that won “Jeopardy” several years back — to help doctors diagnose heart disease.

This means the computing system has been used to cull through patient imaging data, like that from ultrasounds and X-rays, to help doctors more accurately determine if the mass they’re looking at, for instance, is dangerous or just a physical irregularity.

For his part, Cantillon, of Cleveland Clinic, has been working on his own technological solutions for making more efficient diagnoses.

“We currently use an algorithm to perform remote ECG monitoring for hospitalized patients from an offsite central facility — a command center bunker, if you will — of all of our patients at the main campus plus eight other Cleveland Clinic hospitals, including Florida,” he said. “The system vastly outperforms what we were previously using. It wins in every performance category, in addition to being more efficient.”

Cantillon stressed that “this is the future.”

“There’s no putting the genie back into the bottle,” he said. “Patients and doctors will have unprecedented access to health and diagnostic information. However, in the cardiology space, those two major caveats I mentioned need to be kept in mind. Bad information can be more damaging and harmful than no information at all.”

Does the increasing presence of wearables like the Heartsense monitor democratize access to health information, empowering people to feel more in control of their heart health?

Tomaselli says that might be the case, but that, as with most technology, there can be a tendency for people to rely on it a bit too much.

“I think it has made some people be more aware, in specific circumstances, which is always a good thing. However, somebody might become preoccupied with things they might not need to even worry about in the first place,” he said. “We always need to make sure the data is robust. A wearable’s interpretation of data over time might not be a great thing. For example, it could make an interpretation that is not accurate and that may not be something that the patient needs to deal with on an imminent basis.”

In a device-obsessed world, Tomaselli warned that a focus on devices could basically “drive up anxiety” for people and could, in some instances, “increase healthcare costs if these people need to seek out care for their anxieties and stress.”

“I think there is no way to avoid the fact that this is going to be part of medical care going forward. Clinicians just need to think about ways to embrace it in a way that makes sense,” he added.

Looking ahead

Shakur is excited about his device’s impending global launch. What about the company’s next step?

“As a next stage, the company is beginning to look at dangerous life-threatening rhythms such as those that require a defibrillator, to develop a particular algorithm and have more trials and actually look at more sensors to be able to really take monitoring of high-risk patients to the next generation,” he said.

He’s also looking to see how the device could factor in sports-related concussions by analyzing ECG activity for players who have suffered a concussion while playing a contact sport. “We feel this is an under-researched area that’s often neglected, especially in amateur sports,” he added.

In wearable-heavy age, he sees a way for his device to make its mark.

Shakur said, “We wanted to give control back to patients and really allow people to now take onus on their own health in very robust and meaningful way.”

And he hopes the Heartsense monitor can help them do that.

Written by Anik

Leave a Reply